The paper presents a method for the selection of large erratics to be sampled for terrestrial cosmogenic nuclide exposure dating (TCNED) in areas previously covered by Pleistocene ice sheets. Our approach is based on (1) a GIS analysis of an extensive dataset of erratics, (2) field inspection of pre-selected boulders and (3) Schmidt hammer (SH) testing of erratics selected for sampling. An initial database of 491 erratic boulders in NW Poland was filtered using a GIS software, based on their characteristics, digital elevation and surface geology. The secondary data set of pre-selected erratics consisted of 135 boulders \textendash i.e. proper targets for field inspection. Ground-truthing in the field resulted in the final selection of 63 boulders suitable for sampling for TCNED. These erratics are located on moraine plateaux and hills formed during the Saalian glaciation (Marine Isotope Stage 6) as well as Leszno/Brandenburg, Poznań/Frankfurt and Pomeranian Phase ice marginal belts from the Weichselian glaciation (Marine Isotope Stage 2). The GIS desk-based analysis of erratics properties resulted in a 73% reduction of the initial dataset, which demonstrates the added value of this selection technique. The field inspection of pre-selected boulders resulted in a 53% reduction of the number of boulders suitable for TCNED. SH testing of the sampled erratics provided a quantitative proxy of their surface hardness. This allowed the quantification of their weathering degree and identification of erratics potentially affected by postglacial erosion. Our systematic approach to selecting erratics and their SH testing could be a useful tool for other researchers facing the problem of choosing appropriate erratics for TCNED in areas of continental Pleistocene glaciations.